LangChain LLM analytics installation

Setting up analytics starts with installing the PostHog SDK for your language. LLM analytics works best with our Python and Node SDKs.

pip install posthog

Install the LangChain and OpenAI Python SDKs:

pip install langchain openai langchain-openai

In the spot where you make your OpenAI calls, import PostHog, LangChain, and our LangChain CallbackHandler. Initialize PostHog with your project API key and host from your project settings, and pass it to the CallbackHandler.

Optionally, you can provide a user distinct ID, trace ID, PostHog properties, groups, and privacy mode.

from posthog.ai.langchain import CallbackHandler
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from posthog import Posthog

posthog = Posthog(
    "<ph_project_api_key>",
    host="https://us.i.posthog.com"
)

callback_handler = CallbackHandler(
    client=posthog, # This is an optional parameter. If it is not provided, a default client will be used.
    distinct_id="user_123", # optional
    trace_id="trace_456", # optional
    properties={"conversation_id": "abc123"} # optional
    groups={"company": "company_id_in_your_db"} # optional
    privacy_mode=False # optional
)

Note: If you want to capture LLM events anonymously, don't pass a distinct ID to the CallbackHandler. See our docs on anonymous vs identified events to learn more.

When you invoke your chain, pass the callback_handler in the config as part of your callbacks:

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful assistant."),
    ("user", "{input}")
])

model = ChatOpenAI(openai_api_key="your_openai_api_key")
chain = prompt | model

# Execute the chain with the callback handler
response = chain.invoke(
    {"input": "Tell me a joke about programming"},
    config={"callbacks": [callback_handler]}
)

print(response.content)

This automatically captures many properties into PostHog including $ai_input, $ai_input_tokens, $ai_latency, $ai_model, $ai_model_parameters, $ai_output_choices, and $ai_output_tokens. It also automatically creates a trace hierarchy based on how LangChain components are nested.

Before proceeding, let's make sure LLM events are being captured and sent to PostHog. Under LLM analytics, you should see rows of data appear in the Traces and Generations tabs.